Three pillars, one loop
Most "AI in GTM" conversations start with a tool. This framework starts with the organisation: what it exists to deliver, who may decide, which knowledge every project should retrieve, which jobs an agent can hold, and how the whole thing runs without anyone losing track of who owns what. The books are the documents, schemas and configurations that make those answers explicit.
Win the right business.
Account intelligence, outbound, ABM, enterprise deals, pipeline and closing.
Create useful demand.
Positioning, content, search and AI discovery, campaigns, conversion and learning.
Make the promise work.
RevOps, CRM, data, automation, customer research, product adoption and success.
Intent & authority ↓
- Purpose and approved choices guide the work.
- Shelves 01–04 set what the organisation is for, who may decide, where it plays and who owns outcomes.
- Shelf 05 is where demand, closed business and realised value are produced — in three parallel lanes.
Evidence & learning ↑
- Results inform revisions across the system.
- Shelves 06–09 hold the shared brains, the agent contracts, the systems and the runtime that keep execution honest.
- What execution learns flows back up: into the brains, the scorecards and the decisions on the top shelves.
A purpose, an accountable owner, a scope, an approved version, a review date, sources, inputs, expected outputs, permitted actions, acceptance criteria and escalation rules. It can be a document, a spreadsheet, a schema, a workflow or an enforced configuration. A book without an owner is a wish, not a book.
Nine shelves, one question each
Each shelf answers one management question and has a suggested owner. Every book on it is listed below with what the human owns, what the agent contributes and the first artifact to create. Shared GTM books stay visible in every pillar; execution books belong to a lane.
| Shelf | The question it answers | Owner | Books |
|---|---|---|---|
| 01 · Commercial foundations | What value do we create, for whom, and how does it become a viable business? | Founder / commercial lead | 6 |
| 02 · Authority & boundaries | Who may decide, approve and act—and when must the work stop? | Commercial lead + relevant policy owner | 6 |
| 03 · GTM strategy | Where will we play, what will we offer, and which motions deserve investment? | GTM lead with Sales, Marketing and Product | 6 |
| 04 · Management & coordination | Who owns outcomes, and how do people and agents stay coordinated? | GTM leadership + RevOps | 6 |
| 05 · GTM execution | How does work deliver demand, closed business and realised customer value? | The accountable owner of each workflow | 18 |
| 06 · Brains & shared context | Which approved knowledge should every project and agent be able to retrieve? | Named subject owner for each brain | 6 |
| 07 · Agent roles & work contracts | What bounded jobs can agents take on, and what counts as accepted work? | A human owner for every deployed agent | 6 |
| 08 · Systems & integrations | Where are the records, and through which interfaces can work happen? | System owners + RevOps / engineering | 6 |
| 09 · Runtime, assurance & improvement | How does the system run reliably, stay observable and improve under control? | Technical operator + relevant business owner | 6 |
Sixty-six books is the whole organisation, not your first project. The loop section shows how to pick seven.
Commercial foundations
What value do we create, for whom, and how does it become a viable business? Suggested owner: Founder / commercial lead.
Purpose & promise
The commercial outcome the organisation exists to deliver, and the customer experience it commits to.
- Human owns
- Define the purpose, promise and non-negotiables.
- Agent contributes
- Compare plans and outputs with the approved promise.
Customer problems & proof
Describe the problems buyers recognise, their consequences and the evidence that solving them matters.
- Human owns
- Choose which problems deserve attention.
- Agent contributes
- Synthesize interviews and link every claim to its source.
Business model & economics
Explain how revenue is earned and what delivery costs. Separate SaaS subscriptions from service capacity and project margins.
- Human owns
- Set economic assumptions and acceptable trade-offs.
- Agent contributes
- Model scenarios using explicit, reviewable assumptions.
Revenue architecture
Map acquisition, conversion, delivery, retention and expansion into one revenue system with named owners.
- Human owns
- Agree where growth should come from.
- Agent contributes
- Locate gaps, duplicated work and unsupported dependencies.
Buying & value journey
Follow the customer from recognising a problem through buying, receiving value and deciding to renew or expand.
- Human owns
- Validate the journey with real customers.
- Agent contributes
- Connect touchpoints, evidence gaps and handoff failures.
Commercial principles
State how the team competes, communicates and treats customers, including promises it will not make.
- Human owns
- Own the standards and resolve difficult trade-offs.
- Agent contributes
- Check draft communications against documented examples.
Authority & boundaries
Who may decide, approve and act—and when must the work stop? Suggested owner: Commercial lead + relevant policy owner.
Decision rights
Assign an accountable person to decisions about markets, budgets, offers, records and customer commitments.
- Human owns
- Assign decision owners and delegates.
- Agent contributes
- Route decisions to the appropriate owner with evidence.
Agent autonomy
Set authority per action: recommend, collaborate, prepare for approval, or execute within a defined scope.
- Human owns
- Approve scope, limits and stop conditions.
- Agent contributes
- Act only through the permissions available for that task.
Claims & evidence
Link commercial claims to approved evidence, intended audiences, expiry dates and allowed wording.
- Human owns
- Approve new claims and material changes.
- Agent contributes
- Flag unsupported, stale or contradictory claims.
Data & contact rules
Document permitted data use, contact eligibility, suppression rules and the review needed for uncertain cases.
- Human owns
- Approve applicable rules with the responsible specialists.
- Agent contributes
- Apply approved eligibility checks and escalate ambiguity.
Commercial approvals
Define authority for discounts, contracts, paid media budgets, vendors and promises about product or delivery.
- Human owns
- Approve commitments above delegated limits.
- Agent contributes
- Prepare the decision pack and hold the action until authorised.
Quality & escalation
Define acceptance criteria, blocked-work states and escalation paths when evidence, permissions or tools fail.
- Human owns
- Resolve exceptions and accept material outcomes.
- Agent contributes
- Stop, preserve evidence and propose a recovery action.
GTM strategy
Where will we play, what will we offer, and which motions deserve investment? Suggested owner: GTM lead with Sales, Marketing and Product.
ICP & segmentation
Choose the organisations and buying situations worth pursuing. Keep fit, intent and relationship strength distinct.
- Human owns
- Choose segments, exclusions and trade-offs.
- Agent contributes
- Research candidates and show the evidence behind fit.
Positioning & messaging
Define the alternatives buyers compare, the differentiated value and the narrative for each relevant audience.
- Human owns
- Choose the position and approve the narrative.
- Agent contributes
- Compare alternatives and draft evidence-backed variations.
Offers & pricing
Specify the entry offer, value exchange, qualification rules and path to larger commitments.
- Human owns
- Approve packaging, price logic and concessions.
- Agent contributes
- Model scenarios and draft offers within approved rules.
Channels & motions
Choose how customers discover, evaluate and buy. Define each channel's role and the handoffs between motions.
- Human owns
- Choose the mix and investment hypotheses.
- Agent contributes
- Compare performance and surface gaps between channels.
Revenue plan & resources
Translate revenue goals into explicit assumptions about conversion, deal size, timing, retention and delivery capacity.
- Human owns
- Approve targets, budgets and capacity commitments.
- Agent contributes
- Calculate scenarios and expose incompatible assumptions.
GTM readiness & launches
Define the customer evidence, offer, sales readiness, support and instrumentation needed for a launch decision.
- Human owns
- Make launch and market-entry decisions.
- Agent contributes
- Assemble readiness evidence and flag unmet criteria.
Management & coordination
Who owns outcomes, and how do people and agents stay coordinated? Suggested owner: GTM leadership + RevOps.
Roles & ownership
Map human roles, agent mandates and accountable owners for processes, records, metrics and handoffs.
- Human owns
- Assign accountability and resolve ownership gaps.
- Agent contributes
- Check tasks and exceptions have a valid destination.
Programs & priorities
Give every initiative an outcome, scope, owner, budget, dependencies and a decision to continue, change or stop.
- Human owns
- Prioritise initiatives and allocate capacity.
- Agent contributes
- Maintain progress evidence and surface dependency risks.
GTM operating calendar
Connect weekly pipeline reviews, campaign learning, quarter goals, renewals, launches and annual planning.
- Human owns
- Set cadences and choose actions at reviews.
- Agent contributes
- Prepare briefs and flag upcoming decision dates.
Metrics & scorecards
Define the business questions each metric answers, its formula, source, time basis and action threshold.
- Human owns
- Approve meanings, targets and review decisions.
- Agent contributes
- Calculate consistently and separate missing data from zero.
Capacity & dependencies
Connect sales capacity, customer delivery, campaign output and agent workload to realistic commitments.
- Human owns
- Approve staffing and delivery commitments.
- Agent contributes
- Find bottlenecks and calculate the effects of changing load.
Customer learning council
Bring together sales objections, lost deals, adoption issues and customer evidence to revisit GTM choices.
- Human owns
- Decide what changes in product, offer or GTM.
- Agent contributes
- Cluster evidence and distinguish repeated signals from anecdotes.
GTM execution
How does work deliver demand, closed business and realised customer value? Suggested owner: The accountable owner of each workflow.
Three parallel execution lanes. The first listed pillar of a book sets its lane; books tagged with two pillars are shared between them.
Sales lane — Win the right business.
Account intelligence
Build account and buying-committee context from permitted sources, with provenance, unknowns and a reason to act.
- Human owns
- Choose which opportunities deserve attention.
- Agent contributes
- Research accounts and maintain evidence-linked dossiers.
Outbound & social selling
Turn an approved target and point of view into relevant outreach, follow-up tasks and a response-handling process.
- Human owns
- Own relationships, eligibility and external communication.
- Agent contributes
- Draft outreach and prepare tasks within contact rules.
Account-based programs
Coordinate account selection, tailored content, seller activity and buying-group engagement around one account plan.
- Human owns
- Choose account hypotheses and relationship moves.
- Agent contributes
- Prepare account briefs, channel actions and progress evidence.
Discovery & qualification
Establish the buyer's problem, impact, decision path and stakeholder evidence. Treat unknown qualification fields as unknown.
- Human owns
- Conduct discovery and judge buyer commitment.
- Agent contributes
- Prepare questions and map call evidence to qualification gaps.
Deal progression & closing
Use buyer-confirmed milestones, mutual action plans and commercial approvals to move a deal to signature.
- Human owns
- Own negotiation, commitments and close decisions.
- Agent contributes
- Track dependencies and propose evidence-based next actions.
Pipeline & forecasting
Separate stage, close date, forecast category and evidence. Track slippage and compare assumptions with outcomes.
- Human owns
- Own the forecast call and account interventions.
- Agent contributes
- Detect stale deals, date changes and inconsistent evidence.
Marketing lane — Create useful demand.
Demand & allbound
Connect campaigns, events, partners, capture and seller follow-up around a buyer problem and a measured hypothesis.
- Human owns
- Choose the audience, hypothesis and channel investment.
- Agent contributes
- Assemble program assets and monitor agreed handoff signals.
Content & enablement
Build useful buyer content from approved positioning, customer questions and evidence, then adapt it for the buying journey.
- Human owns
- Choose the editorial point of view and approve publication.
- Agent contributes
- Research, draft, repurpose and check sources and voice.
Search & AI discovery
Improve discoverability and accurate representation in search and AI answers. Record the query, engine, date and evidence for observations.
- Human owns
- Choose priority topics and approve public claims.
- Agent contributes
- Audit content and run repeatable visibility observations.
Paid media & experiments
Connect creative, audience, offer and conversion event to a test with a budget, duration and decision rule.
- Human owns
- Approve spend, targeting and material campaign changes.
- Agent contributes
- Draft variants and report performance against test rules.
Conversion & lifecycle
Improve the path from interest to a useful next step across pages, forms, nurture and routing.
- Human owns
- Approve experience changes and contact policies.
- Agent contributes
- Find drop-offs and propose measurable experiments.
Attribution & learning
Define attribution windows and assumptions, join campaigns to outcomes and distinguish observed credit from causal evidence.
- Human owns
- Choose the model and interpret its limitations.
- Agent contributes
- Reconcile touchpoints and report results with uncertainty.
Operations & Product lane — Make the promise work.
CRM & revenue operations
Define object models, lifecycle states, required fields, ownership, handoffs and the process those records support.
- Human owns
- Approve the operating model and significant record changes.
- Agent contributes
- Audit hygiene and prepare controlled changes and task queues.
Data & analytics
Create usable datasets with explicit grain, identifiers, history, definitions, freshness and quality checks.
- Human owns
- Approve authoritative sources and acceptable quality.
- Agent contributes
- Reconcile records and identify measurement gaps.
Workflow automation
Specify triggers, eligibility, actions, duplicate prevention, errors and acceptance criteria before automating a process.
- Human owns
- Approve the process and permitted effects.
- Agent contributes
- Prepare changes and execute authorised workflow steps.
Customer research & PMF
Collect customer evidence about value, unmet needs, alternatives and willingness to change; retain contrary evidence.
- Human owns
- Choose research questions and interpret product implications.
- Agent contributes
- Code interviews and connect patterns to their original sources.
Product adoption & PLG
Connect activation and product usage to customer value, sales-assist triggers and product improvement hypotheses.
- Human owns
- Define value milestones and approve product changes.
- Agent contributes
- Monitor cohorts and draft evidence-based improvement proposals.
Customer success & expansion
Make the sales promise operational through onboarding, value reviews, renewal planning and evidence-based expansion.
- Human owns
- Own the relationship and renewal or expansion commitment.
- Agent contributes
- Prepare success reviews and surface value-delivery gaps.
Brains & shared context
Which approved knowledge should every project and agent be able to retrieve? Suggested owner: Named subject owner for each brain.
ICP & market brain
Store approved segmentation rules, market hypotheses, competitive evidence and known exclusions with versions and review dates.
- Human owns
- Approve segmentation and market interpretations.
- Agent contributes
- Retrieve the applicable version and propose updates with evidence.
Brand & content brain
Keep positioning, audience narratives, claim sources, style examples and reusable approved assets in one maintained context pack.
- Human owns
- Approve voice, narrative and claim changes.
- Agent contributes
- Retrieve relevant examples and check drafts for consistency.
Account & deal memory
Keep account facts, stakeholder roles, decision history and current commitments distinct from assumptions and draft recommendations.
- Human owns
- Confirm relationship judgments and commitments.
- Agent contributes
- Summarise evidence while retaining sources and timestamps.
Product & customer brain
Maintain product truth, service constraints, pricing rules, customer questions and evidence of delivered value.
- Human owns
- Approve product promises and delivery constraints.
- Agent contributes
- Retrieve current facts and flag contradictory or stale material.
Metrics & definitions brain
Version metric formulas, lifecycle definitions, time windows, units, exclusions and canonical data sources.
- Human owns
- Approve changes to business meaning.
- Agent contributes
- Use the correct contract and explain comparability limits.
Decisions & experiment memory
Record what was tried, what happened, why a decision was made and what evidence would justify revisiting it.
- Human owns
- Accept learnings and authorise changes to the playbook.
- Agent contributes
- Retrieve relevant precedents and propose supported updates.
Agent roles & work contracts
What bounded jobs can agents take on, and what counts as accepted work? Suggested owner: A human owner for every deployed agent.
Research & intelligence agent
A proposed role for sourced account, market and competitor research; outputs include evidence, freshness and unresolved questions.
- Human owns
- Set the research question and accept evidence quality.
- Agent contributes
- Produce structured dossiers through approved sources and tools.
Campaign & content agent
A proposed role for briefs, drafts, content checks and campaign preparation using the current brand and ICP brains.
- Human owns
- Own narrative, budget and publication decisions.
- Agent contributes
- Produce reviewable assets and a clear change summary.
Deal & forecast agent
A proposed role for deal preparation, milestone tracking and pipeline anomalies, with buyer evidence kept separate from model inference.
- Human owns
- Own relationships, deal judgment and forecast commitment.
- Agent contributes
- Prepare evidence-led reviews and suggested next steps.
RevOps & data agent
A proposed role for data checks, process diagnostics, dashboard calculations and controlled workflow or record updates.
- Human owns
- Approve definitions, process changes and write scopes.
- Agent contributes
- Produce diagnostics, change previews and execution receipts.
Product & customer agent
A proposed role for interview synthesis, adoption patterns, customer health evidence and product feedback aggregation.
- Human owns
- Interpret customer context and make product or renewal decisions.
- Agent contributes
- Link patterns and recommendations to customer evidence.
Coordinator & quality agent
A proposed role for preparing tasks, selecting approved context, coordinating handoffs and checking acceptance criteria.
- Human owns
- Own priorities, acceptance and exception decisions.
- Agent contributes
- Coordinate bounded work without granting itself new authority.
Systems & integrations
Where are the records, and through which interfaces can work happen? Suggested owner: System owners + RevOps / engineering.
CRM records & objects
Map companies, contacts, deals, activities and custom objects to canonical owners, identifiers and controlled write rules.
- Human owns
- Choose authoritative records and approve schema changes.
- Agent contributes
- Read and update only permitted objects and fields.
Engagement channels
Register the available email, calendar, advertising and messaging interfaces, including account permissions and operational constraints.
- Human owns
- Own accounts, channel policy and external communication.
- Agent contributes
- Prepare or execute only the actions supported and authorised.
Website & measurement
Connect content, forms and analytics to defined events, campaign identifiers and the customer journey.
- Human owns
- Approve public changes and measurement definitions.
- Agent contributes
- Prepare content changes and check event consistency.
Product, billing & service
Connect usage, delivery, contracts and billing so customer value and commercial reporting can be reconciled.
- Human owns
- Resolve which source is authoritative for each measure.
- Agent contributes
- Join authorised signals and surface mismatches.
APIs, MCP & tool contracts
Describe each callable capability, input, output, side effect, access scope and failure behaviour; connectors are implementation choices.
- Human owns
- Approve capabilities and access scopes.
- Agent contributes
- Call the appropriate tool and retain its result and action receipt.
Identity, events & sync
Define stable IDs, timestamps, event schemas, deduplication and reconciliation across the GTM stack.
- Human owns
- Approve mappings, conflict rules and source priorities.
- Agent contributes
- Reconcile changes and prevent repeated effects.
Runtime, assurance & improvement
How does the system run reliably, stay observable and improve under control? Suggested owner: Technical operator + relevant business owner.
Identity & permissions
Give each runtime a scoped identity, managed secrets and enforceable permissions linked to its work contract.
- Human owns
- Approve access and review changes.
- Agent contributes
- Operate within the granted identity and report denied actions.
Runtimes & schedules
Specify where work runs, what triggers it and how queues, retries, timeouts and concurrent tasks behave.
- Human owns
- Approve service levels and workload limits.
- Agent contributes
- Execute eligible work and checkpoint progress.
Storage & retrieval
Map documents, structured records, logs and indexes. Retrieve authorised context with provenance, freshness and version boundaries.
- Human owns
- Choose canonical stores, retention and retrieval rules.
- Agent contributes
- Retrieve relevant context and identify missing or stale evidence.
Observability & cost
Track business outcomes separately from execution traces, quality, latency, tool failures and spend.
- Human owns
- Set thresholds and decide interventions.
- Agent contributes
- Collect receipts and flag quality, cost or performance deviations.
Evaluations & releases
Use representative GTM cases to check evidence, calculations, tool behaviour and usefulness before releasing changes.
- Human owns
- Approve acceptance thresholds and releases.
- Agent contributes
- Run evaluations and report failures without rewriting the standard.
Recovery & retirement
Define pause controls, rollback, incident handling, restore procedures and how an agent or workflow is retired.
- Human owns
- Own incident decisions and service restoration.
- Agent contributes
- Preserve state and execute the authorised recovery procedure.
Build one useful loop first
The shelves describe the organisation. Your first implementation needs only the books that support one real commercial outcome. Three steps, then expand.
- Choose the outcome. Start with a revenue problem — poorly qualified accounts, inconsistent follow-up, unreliable pipeline data, slow customer activation. Record the current baseline before anything changes.
- Prepare the minimum books. Open the relevant books and add them to a build list. Assign owners, create the first documents, define permitted actions and acceptance criteria.
- Run, review, improve. Trial the workflow on a real case with approved tools and representative data. Review output quality, commercial usefulness, cost and exceptions before expanding.
Three worked projects
Each one is seven books used in sequence, with the outcome it produces and the measure that tells you whether it worked.
Launch an ABM program
Outcome: A qualified account shortlist, approved narrative and coordinated account actions.
Build a content-to-demand loop
Outcome: A sourced content asset connected to discovery, conversion and commercial follow-up.
Repair the pipeline system
Outcome: A documented pipeline issue, approved change and measurable follow-up.
The interactive bookshelf lets you filter by pillar, load a project, add books to a build list and export their implementation prompts as markdown — one file per book, or the complete guide.
Four autonomy levels, set per action
Autonomy is not a property of an agent; it is a property of each action the agent may take. The Agent autonomy book on shelf 02 records one level per action in an authority matrix, with stop conditions. Most first implementations live on the first two rungs.
An agent acts only through the permissions available for that task, and never grants itself new authority. When evidence, permissions or tools fail, it stops, preserves the evidence and proposes a recovery action — that is the Quality & escalation book, and it is on the shelf before any execution book for a reason.
How the shelves map to the Revenue Puzzles operating model
The five layers on this site — Business, Intelligence, Work, Execution, Output — are the same organisation seen from the outside. This is where each layer lives on the shelves.
| Layer | Where it appears in the framework |
|---|---|
| Business | Commercial foundations, GTM strategy and the three execution lanes (shelves 01, 03, 05). |
| Intelligence | Shared brains, definitions, evidence and decision memory (shelf 06). |
| Work | Management, programs, human ownership and agent work contracts (shelves 04, 07). |
| Execution | Approved workflows, integrations, tools and runtime controls (shelves 08, 09). |
| Output | Accepted artifacts, recorded actions, decisions and measured business outcomes — governed by shelf 02. |
Two books already have working instruments on this site: Metrics & scorecards and the Metrics & definitions brain are what the GTM Control Room implements; the GTM operating calendar is the Sales Leaders Calendar. The CRM & revenue operations and Deal progression books are expanded in the CRM Revenue Growth and Enterprise Sales playbooks.
How to use it — and what it is not
- Not a maturity model. Nothing here scores your organisation; it names what has to exist for an agent to do useful, bounded work.
- Not a tool list. Connectors, models and platforms are implementation choices recorded in the APIs, MCP & tool contracts book, not framework decisions.
- Not sixty-six documents on day one. Seven books and one real case, then the next loop.
- The agent roles on shelf 07 are proposed roles. Each one exists only once a human owner has signed its work contract.
The GTM Agentic Organisation, framework v1.0 — Revenue Puzzles working frameworks. The book format is: id, title, pillars, search terms, purpose, first artifact, human role, agent role. The interactive bookshelf and the complete guide are generated from the same list, so the three stay in step.
The first session is a revenue problem and a baseline, not a tool. Tell me what is going on and I will say which seven books come first.